Display Monitor White vs Color Brightness

[MrAl]

Hello there,

I have noticed a problem that seems to plague computer monitors and maybe TV sets.

First, most computer monitors work by showing three different colors: Red, Green, and Blue, which i will abbreviate as R,G,B as needed, for each pixel. The monitor achieves various colors and shades of those colors by varying the level of white light that is allowed through the aperture for that color. If the pixel in a certain location needs to show red, it will open the red wider and the other colors will remain closed. The viewer then sees red for that pixel. For violet, it would open both the red and blue and keep the green closed. For white, it would open all three and not keep any closed.
That is the basic operation and it achieves the aperture change in various ways.

Each pixel color R, G or B emits a certain amount of light when the aperture is open for that color. It's like a tiny flashlight that gets turned on, and there is one for each of the three colors. The more each one emits, the brighter the screen appears.

The problem is since there is the equivalent of three flashlights behind the screen, if one is turned on it only emits about 1/2 of the light that gets emitted when two are turned on, and only about 1/3 of the total light output that gets emitted when all three are turned on which is what has to happen to create white.
To put it another way, when white is emitted the screen outputs THREE times the light intensity that it does for any individual color like blue. This of course means that anything that is white in the picture is going to be brighter than anything of any one or two colors. This is especially true when one color is displayed across a background of white. The color gets drowned out by the white surrounding the color picture. This is true of many types of graphics too like a program that shows a file listing, which is made up of mostly dark text on a pure white background. The white background is almost blinding.

Ok white is almost blinding, so just turn down the brightness right?
I wish that would work. What happens is all the colors decrease in brightness as well.

What this means is that if you have the monitor set for viewing that file listing on the white background so the background doesnt blind you then when you go to look at a picture of a natural scene or human subject they appear far, far too dark.

This problem is not subtle either. There is a marked difference in brightness of a white screen compared to a single color screen like blue.

This whole idea comes into play when converting color photos to black and white, because each intensity level has to be made to be the average of the three colors:
C=(R+G+B)/3

(Note there is a small correction factor for the three colors, but it is very small compared to the issue being discussed here so it is left out for simplicity).

What that formula does for three levels of the same high intensity is it makes the brightness of the black and white picture the highest pixel color which is 255 on most monitors. This happens because R, G, and B are all 255, so the average is 255 also:
C=(255+255+255)/3=255

So the pixel for that location would be 255. There are still three colors however, so each one would be made 255.

So what happens when a single color is displayed vs when white is displayed?
For a single color the maximum intensity setting is 255, so say blue would be 255.
But for white, we would have the total sum 255+255+255=765.
So we would be staring at a pixel with relative intensity of 756 vs a pixel with intensity of only 255. That's the problem. The monitor or graphics card does not average out the three as we do when creating a black and white photo from a color photo. If it did, and if the colors were really of equal intensity, the right levels would not be 255, 255,255 but would be 255/3, 255/3, 255/3, for a total intensity level of 255 again. So both white and blue would show up (ideally) at the same intensity and so the colors would appear normal compared to white.

So the question is, does anyone have any idea what to do about this?
What it leads to is you have to adjust the brightness for reading text on a white background and readjust the brightness for viewing photographs.
It's hard to do this back and forth when say surfing the web where you have to read and view photos sometimes.

BTW this has been a known issue for projectors in the past. Epson has apparently improved their projectors to eliminate this problem. It seems to appear in monitors too though.

 

[ronsimpson]

That's the way it has always been.

Walk outside (day time). Assume that the sun give a light level of 765. (visible light only) {using your numbers}
Hold up a red filter so you only see red. The light level is 255. (because you filtered out 2/3 of the light)
White has 3x the energy as blue because that's the way it works. White is 1/3+1/3+1/3.
The original system was built to go from an image to camera to broadcast system to TV set to eye. It works.
A computer can make pictures that would not be good.

So both white and blue would show up (ideally) at the same intensity and so the colors would appear normal compared to white.

No! For B to have as much energy as (R+G+B)....well.....that goes against simple math.
If you could make blue 3x brighter, and red 3x brighter and green 3x brighter than white will be 3x brighter.
I am looking at a screen with a blue-gray background with black text. The background is not full white, for a reason.
Don't complain to TV and monitor makers. Complain to WEB developers. They have control of what you see. How bright and what color.

Epson has apparently improved their projectors to eliminate this problem.

Improvement? Some monitors (TVs and such) have a maximum amount of power they can output. I am thinking of CRT days. If the high voltage section can only generate 40 watts then..... probably R=25 watts or G=25 watts or B=25 watts and R+G=40 watts and R+AG+B=40 watts. This causes some bad effects that get monitors rejected.

 

[Nigel Goodwin]

BTW this has been a known issue for projectors in the past. Epson has apparently improved their projectors to eliminate this problem. It seems to appear in monitors too though.

I've never heard of any such suggestion, or seen any such effect - obviously full brightness white is brighter than any full brightness single colour - but the real world is like that as well.

Your problem is more likely to do with the source material, although I can't say I've ever found any such problems.

 

[MrAl]

Hello again,

Well, your replies are interesting but this is a 'problem' whether or not it is recognized yet in the industry or not. The problem exists across different sources too as i looked into that. It doesnt matter if i view a picture on my computer already or on the web in a different application. I will say however that if we had control over each application rather than globally (adjusting monitor settings instead of settings in the web browser and picture viewer for example) it would allow setting the brightness level for each application which assuming it is used for a certain purpose it would at least offer some relief. Let me try to show an example or two and explain this a little better.

First, the graphics card is sent three digital data bytes, one for each color R,G, and B. The bytes have values 0 to 255. This is standard but there are other standards which i wont get into right now.
For a single color we have a value of 255, and for white we have three values of 255 which means white comes out brighter by a factor of 3. Yes, there is a correction factor for the three different colors for example green is seen as brighter than red so the correction factor for green is a little less than 1. But it's not enough to make a difference for this discussion because we're talking a 3 to 1 ratio not 1.2 to 1. So this issue is much more profound than the human eye perception correction. BTW it does not matter as much what the physical power levels of each color are because the monitor is not nature, it is an imitation of nature and needs to be dealt with at a different level. Thus although blue might have the most energy in nature, green is perceived as brightest on the monitor.

So with that in mind, we have a color blue sent to the graphics card, at a level of 255 (the max). The intensity is about 255. Any pixels that are blue have a level of 255.
We have another pixel that is blue, but only at intensity 128. This second pixel (ideally) shows up as half as bright as the first. No problems so far, and this is roughly true of either of the three individual colors.

Now we want to display violet. This requires both blue and red. So we simply send two channel data to the graphics card: blue 255 and red 255 for max violet, and for the second pixel blue 128 and red 128. The first pixel has total intensity 510 while the second 256, which is roughly 1/2 as bright as the first, and all is well? This would be fine except for the fact that the pixels in the first example, the blue ones, will show up as 255 and 128, which is only half as bright respectively as the ones in this new set.
So we see blue as half as bright as violet because violet requires two channels.

What to do?

Well, the graphics card (or monitor) has all the information it needs to compensate, we would just need to 'filter' the color channels. This could be viewed as a three dimensional filter, not a single dimensional filter like we are used to seeing in electronic signal processing like a bandpass filter or low pass filter.
Lets look at the two sets of data above in light of this 'filter'.

First we have the blue pixel at 255, and another at 128. We need the second to come in at 1/2 intensity (or nearly that). So the first rule is when we get a single channel of data we feed it through without attenuation. This means we display 255 and 128 respectively for blue.
Next we have violet which requires *data* of R=255 and B=255 and the second pixel is R=128 and B=128. But we want the intensity levels to be the same for these two pixels (respectively) as the first two pixels of blue only, so the second rule is when we have two channels of data we have to divide each by 2. Dividing R/2 and B/2 we get 127 and 127 for the R and B channels for the first pixel, and we get 64 and 64 for the second set, so for the first violet pixel we get a total intensity of 254 and for the second violet pixel we get 128, which now is of the same intensity as the blue pixels respectively.
Next we have the all encompassing white, which requires all three colors RGB. When we get all three channels the rule is we have to divide each channel by 3. Thus when we get R=255, G=255, and B=255, we divide each by 3 and so the total intensity is again 255 for this pixel, which again is the same as both the violet and the blue above.
Thus ideally we get the same brightness for ANY color or white.

As i said before the RGB channels might also have to be filtered to reflect the way the human eye responds to the different color frequencies, but i dont care about that right now because that influence is much less than what we are talking about here now.
Also, the above may not be the exact method to compensate for this 'super' white intensity problem, it is just a quick method to show something along the lines that could be done. Obviously Epson has found a way to compensate in their projectors so there must be a way to do it for monitors and/or graphics cards. I am just so surprised that nothing has been done to date that i know of.

To illustrate the problem better, imagine viewing a full color picture with several colors like red, blue, green, orange, violet, yellow like a basket of fruit, and one a white background. The human eye sees this in real life with no problem. On the computer however the colors come out noticeably darker than the white background so it is harder to make out detail of the colored objects while the white is super bright.
If you look at the same picture with color levels filtered the right way, you can easily see the fruit without the white blinding you.
In a text application with white background vs a full color image of nature, if you adjust the white level with the brightness control for a comfortable level for reading then when you go to look at a photo the photo seems too dark. If you adjust for the photo, then the white background for the text seems too bright.

I dont remember having this problem with a CRT monitor, but maybe i did, as that was a long time ago since i had one of those for the computer.

If you would like to review the Epson ideas on this just do a search for the Epson projectors and color content or something like that.
Also note that most monitor calibration routines require only getting the colors themselves right, not the relative levels of brightness.
Also note that when a monitor is tested for brightness the white level is the only thing that is measured. Thus if a manufacturer makes a bright white they get brownie points for having a bright monitor, even though the colors may not be as bright.
They also talk about this a little as "color brightness" vs white brightness.

Perhaps some ideas how to deal with this?
Right now the only solution seems to be to change the brightness depending on what is being viewed, which is certainly a pain in the neck.
One thing that would help is an icon that can be clicked to set the monitor to 'level photo' and another to click to set to 'level text' for example, if that is possible.

Any other ideas?

 

[Nigel Goodwin]

Perhaps some ideas how to deal with this?

I'm still totally baffled by your supposed 'problem'? - it sounds like it's working exactly as it's supposed to, and identically to 'real life'.

A pure red (or other) colour isn't as bright as pure white, it's not supposed to be - and it's not so in 'real life' either - stick a white light through a prism and it's shown to be a combination of other colours (basically the three primary ones) making the white up.

 

[ronsimpson]

I dont remember having this problem with a CRT monitor, but maybe i did

A LCD monitor starts out with white light and removes what you don't want to see.
A CRT monitor starts out with black (no light) and makes what you are to see.
The end effect is the same. CRT or LCD

The people that make content (what you read on the internet), they choose color and brightness. They are who you should be unhappy with. If some one is uneducated enough to make a page with full white background with blue letters, then your eyes will hurt.

The computer + monitor can only do what it is told to do. Any effort to alter the video will have bad effects.

This is like demanding a car that has the power to pass a truck on the road. Then complaining that it can/will go over the speed limit. Don't drive at full power in a school zone. Don't say the car should know it is a school zone. Tell the driver to back off. I understand you are just the passenger in the car.

 

[MrAl]

Hi,

I appreciate you guy's inputs. It is good to hear what other people are thinking about this.

First let me say that the monitor is not 'real life'. It is not the same. To be more accurate, the human experience with a computer monitor is not the same as with real life. Colors on a monitor must be different than in real life in order for the human operator to get the impression they are dealing with real life objects. From what i have found it looks like they need to be brighter than in real life.

Second, i dont believe it has as much to do with the source as it does with the basic intensity difference between a single color and white. That's because i can find many sources that do not work as they should when the monitor is adjusted for one type of viewing as opposed to another type of viewing. The two types are roughly photos and text, but text with a white background, and we have no control over some sources that have text and a white background. We also dont have any control over what the web designer puts on the web site, but if they are doing something wrong then they must all be making the same mistake.

If you dont agree i can go further with this, but maybe we should just put the differences aside for a moment and ask the question what can be done about this. Even if i happen to be the ONLY one having this problem, then what can i do about it? Here's what i would like to see happen...

Turn on the computer, read text on a white background without too much eye sore. I might have to read for an hour, two hours, maybe more, but can take rests in between.
While i am doing that, i may want to refer to some drawings. If the drawings are schematics with white background and black lines, then they may appear the same as the text, so no problem yet. But the drawings may include color, and maybe even somebody sends me a picture of them in the email so when i check i want to see the picture, without having to adjust for brightness all over again.
That's what i would like to do, regardless what anyone else is doing or posting on the web. Of course i dont expect to see extreme examples like very dark pictures, or even very light pictures. I would have to make an exception there. But there is a problem here and i need a solution.

Right now if i adjust the brightness for viewing photos, i can see most on my computer and most on the web without too much difficulty. I'd say min 80 percent come in good. But then if i go to read black text on a white background (pdf's etc.) the white is so overwhelming that i feel the need to put on sunglasses.

So i need a solution, if there is any.

Thanks a bunch

 

[Nigel Goodwin]

without having to adjust for brightness all over again.

I've NEVER adjusted brightness on a monitor once it's been installed and adjusted to my preference - which again is why I'm totally bemused by your 'problem'.

Do you have some sort of 'auto-brightness' set to ON?, or some kind of weird picture processing?.

Failing that it sounds like you perhaps have the contrast much too high?.

 

[ronsimpson]

I see what you say. My PDF printer, for making PDFs, forces the background to white and the foreground to black. No choice. I can not choose a gray page color. The PDF reader also does not have a way to turn down the brightness (actually contrast).

I have CAD programs that let me set the "contrast" for what I see on the monitor differently than what I get from the printer.
On the printer I want white (no ink) for the entire page. On the monitor I might want a black background or gray or tan, (do not want to print like that).
PDF was developed for printing where the background is white.

On monitors and TVs, brightness and contrast are confusing to the user. Contrast sets the distance between black and white. (normally used to set white lever) Brightness moves both black and white together. (normally used to set black to very close to 0 output)

 

[MrAl]

Hello again,

Nigel:
What you said about the contrast seems to be part of the issue. That makes sense. If i turn the contrast down, way down, then i can turn the brightness up and strike a balance between reading text and viewing photos. So what it looks like is i will have to settle for some in between setting of contrast and brightness if i dont want to have to adjust the brightness with each application.

It's a shame though because when i turn the contrast down i loose part of the black level...the level that makes blacks pure black, and that bothers the contrast of both text and pictures, although i think it bothers pictures more. It seems like the only way to deal with this though.
So i guess i'll have to adjust the two very carefully and do some experimenting to try to get a balance between the two basic types of applications.

When i first looked at turning the contrast down it didnt seem like that was a good idea because the whole monitor seems to take on a sort of glow or overcast. But now it is looking like there has to be at least some contrast reduction. Too bad too as this is supposed to be a "high contrast" monitor.
Name is Samsung, but made in China anyway

Thanks for the ideas and suggestions.

 

[ronsimpson]

Here is a test pattern. You should adjust brightness so you can see 5% or at least 10% brightness.
Then adjust contrast so 95% can be seen, and you are happy with eye strain.
I have many of test patterns if any one wants them.

 

[MrAl]

Hello again,

Ron:
Oh good, thanks for the test pattern. I'll have to try using that for the adjustments as that was what Nigel was pointing too as the main culprit and it does look like that has a lot to do with it with my monitor.
A question though...what do you mean adjust for 5 percent brightness, then 95 percent, i am not sure what you mean by that.
Cool looking pattern too

 

[ronsimpson]

MrAl,
There are areas from black (0%) to white (100%), and squares for 10, 20, 30....90, 100%.
The pattern was designed for CRT monitors.
You want to get all the information.
>It is very possible to set the brightness so all numbers from 0 to 30% look like 0% You are missing many data points.
>It is possible to set the contrast so 80% through 100% look like 100%. Missing data.
You want to see that 0% and 10% are different. If you can so you can see the difference between 0% and 5%.
You want to see that 90 and 100 are different.
Using a 8 bit system. You want to see all 256 values.

I have done too much work with medical monitors where doctors are looking for a tumor. The tumor is only slightly different than good stuff. It is very important to see all the data. If you are only chatting on the web........it probably does not matter much. lol

 

[MrAl]

Hello again Ron,

Well, thanks for the explanation, however in that case the test pattern looks faulty.
For example, the 10 percent looks like the 20 percent, looks like the 30 percent, so something is wrong with the test pattern.
I suppose i could correct it, but see if you can find the original first.

Taking another look it looks like maybe they are addressing adjacent squares rather than the named squares themselves.
It looks like i can adjust it going by that.

Interestingly, i just found out another point while looking at the adjusting software. Something they are calling "Presets".
It looks like it may be possible to create a Preset for different applications after all. If this works right, i'd be able to set higher contrast for some things and lower contrast for other things. I'll have to look into this a little more yet though. They are saying it can be set up per application, or hot key, or in the task bar or in control panel. Something there should work.
After adjusting a little bit it seems better though, so i'll try again and see how good it can be.

Thanks again.

 

[ronsimpson]

The "20%" text is on an adjacent square. The 20% square can not have text on it because that is where a light meter goes, and text will effect the reading. If you are adjusting by eye it does not matter.

You can also look at the "wedges" Where the light slowly changes from 0 to 100%.

 

[MrAl]

Hi again,

Yes i noticed that too now, and how useful this kind of pattern can be.
I adjusted and am getting a little better results, although i do find that individual application adjustments works wonders. Problem is, the "Presets" for the high end AMD graphics card i have DO NOT WORK. Creating a preset does nothing for anything on the screen. What a shame, or should i say what a sham. Over the last few years i have come to hate AMD and would do anything not to use anything every made by them or touched by their hands. I've been so disappointed too many times. The AMD 8 core processors (ALL OF THEM) are really 8 integer cores plus 4 floating point units rather than 8 full cores that would be 8 integer cores plus 8 floating point units. I found out too late after i bought one. I read that they hate to trade off real estate for proper heat dissipation, but im just a user i dont care what they did it should have been a full 8 core unit because it was sold as an 8 core unit, not a 'broken' 8 core unit. Their plans are to keep it this way well into the future too even with core improvements. The Phenom core was a true (up to) 6 core unit, after that it was all downhill. Now it's hard to get one.
But anyway, back to the monitor...

Maybe Nigel can give us a little background info on how the LED's in an LED backlit monitor are supposed to work, as to their type, placement, etc. Is it possible to modify the driver circuit to lower the LED light itself? This would give me a custom intensity setting i could set up with my own software and a little hardware to switch automatically when i go to a different application. I wonder what kind of driver they use, since it is all low voltage LED stuff. The power supply is a 14vdc 2.5 amp wall wart

 

[ronsimpson]

There are many different types of monitors.
If you have a LED back light then they are using white LEDs running from a constant current power supply. Most likely it is adjustable from some control. Probably brightness.

There are several things that sets what you see.
There is a digital number (maybe 0 to 255) that goes into the monitor for each color. Could be 10 bit or 12 bit.
In the high end monitors there is a lookup table.
>If the table is "zeroed" then the table does nothing. What goes in comes out.
>The table can be filled in to give you gamma correction.
**broken link removed**
Next there is a way to set how "open" the LCD shutter is at zero. This should be set to compensate for room brightness. In a very bright room there needs to be much light at zero. Then if you turn off the lights in the room the monitor will seem to glow at zero. So the brightness needs to be set much lower for a dark room.

Brightness sets how open the LCD shutter is at Full white.

I hate to tell you this but after the lookup table the digital numbers are converted to analog.
Brightness is "offset" or how much voltage you want applied to the LCD shutter when "zero" is sent from the computer.
The Contrast sets the gain of the analog amplifier. It sets the distance from black to white.

There are several ways to accomplish this. Hope this helps.
------------------------------------------------------
>Brightness may drive back light current directly. But if the current is too low the color of white will shift to slightly yellow.
>You can keep the LED current constant and adjust the LCD voltage to reduce the brightness. But the DAC may only have 10 bits and as you change brightness greatly you loose how many levels of light output you have. Example: with a 10 bit DAC, running at 1/2 brightness you really only have 9 bits of levels. At 1/3 brightness you are down to 8.4 bits.
------------------------------------------------------
This is probably too much information.
For printing/pictures the monitor should match the paper.
For web surfing the monitor should make you eye happy.
For medical monitors, (complicated) the monitor is adjusted to give the most information. (not a real world picture)

 

[MrAl]

Hello again,

Thanks for the detailed explanation of those things.

What i am dealing with right now is the monitor as a whole, that actually has three contrast controls and three brightness controls. Either of them adjusts those things, but one of the brightness controls works best (maybe two of them) because it does not cause a yellowing with a decrease in brightness.

But any of these requires a physical action on the part of the user in one of three ways:
1. Push buttons on the bottom of the monitor to change settings like brightness and contrast.
2. Open an application that allows the user to set brightness and contrast.
3. Open same application but different window to set brightness and contrast.

The three ways are slightly different, but what they have in common is that i have to do something either with the mouse which takes several clicks, or with the buttons on the monitor. That's what i do now (well, in the application).

What i would like to do is have this happen automatically at best, but also acceptable would be to click an icon to change the two settings with one click. It would go like this:
Click icon 1 for high brightness.
Click icon 2 to go back to low brightness (reading text).

Or something like that. The "Presets" were supposed to allow that, but they dont work. So i would have to implement this myself, by i guess, altering the current to the LEDs. This would probably pass for good enough. So i'd have to have some idea about the best way to reduce the current to the LEDs in a backlit LED monitor. I would hope i would not have to dig into the circuit too deeply, but who knows.

Thanks again for all the information you have provided thus far.

 

[ronsimpson]

I feel your pain. I just got a bead room TV. LCD
In the daytime it is not bright enough. At night it is blinding.
Last night I was watching a dark program which was interrupted by a commercial with a white background and small black letters. Many of the commercials are very white, often flashing white. Some people complain the commercials are loud. The are also bright.